Deep Recurrent Optimal Stopping

Authors: Niranjan Damera Venkata, Chiranjib Bhattacharyya

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 5 Experiments We compare our OSPG algorithm using deep neural networks (DNN-OSPG) and recurrent neural networks (RNN-OSPG) against the following model-free discrete-time optimal stopping approaches.
Researcher Affiliation Collaboration Niranjan Damera Venkata Digital and Transformation Organization HP Inc., Chennai, India niranjan.damera.venkata@hp.com Chiranjib Bhattacharyya Dept. of CSA and RBCCPS Indian Institute of Science, Bangalore, India chiru@iisc.ac.in
Pseudocode Yes Algorithm 1 Pseudocode for mini-batch computation of our temporal OSPG loss
Open Source Code No The paper does not provide any explicit statements or links indicating that the source code for the methodology described is publicly available.
Open Datasets Yes We select 17 multi-class time-series classification datasets from the UCR time-series repository [11] (see Appendix D for details of the datasets and selection process).
Dataset Splits Yes We train models on ten random 50% train-test splits, holding 20% of training data as a validation dataset.
Hardware Specification Yes All experiments were performed on a shared server configured with 2 Intel Xeon Silver 12core, 2.10 GHz CPUs with 256GB RAM and equipped with 6 NVIDIA 2080Ti GPUs. However, experiments were run on a single GPU at a time and no computation was distributed across GPUs.
Software Dependencies No The paper mentions software like Keras and Adam but does not specify their version numbers or the version of Python used, which are necessary for reproducible software dependencies.
Experiment Setup Yes Table 2 shows general hyper-parameter settings used for all experiments.